Skip to content

Command Console

Lennox98 edited this page Sep 6, 2026 · 1 revision

Command Console

CommandDeck exposes an allowlisted command language for local casework. It is not a shell, a browser automation layer, or a platform-control API. Commands operate against the selected local case and return a normalized command, structured data, evidence links, confidence where available, and suggested follow-up commands.

Available Commands

Area Commands
Context /help, /context, /history
Search /find posts, /find accounts, /find mentions, /find links
Sources /source add
Profiles /profile show, /profile history, /profile connections, /profile activity, /profile aliases, /profile compare
Network /graph build, /graph centrality, /graph communities, /graph common, /graph path
Timeline /timeline build, /timeline compare
Casework /case create, /case status, /case note, /case tag
Review /review queue, /review approve, /review reject, /confidence set, /contradictions, /duplicates, /gaps
Runtime /agent status, /queue
Export /report generate, /export case

Use --case <id> with a supported command to target another local case.

/find posts --query "term" --min-risk 45 --limit 20
/profile connections @account
/graph path @account_a @account_b --max-hops 4
/timeline build
/report generate --format pdf

The versioned source of truth is docs/COMMANDS.md. Commands from older concept catalogs that are not listed there return not_available until an explicit, tested handler is added.

API Entry Point

The browser sends commands to POST /api/commands. On loopback this follows the configured local security boundary. If an API token is configured, the route requires that token. A network bind additionally requires explicit opt-in and a Host allowlist.

See Security and Trust before enabling any network-facing configuration.

Clone this wiki locally